Visualization + Analysis Blockchains Are Networks Time-series - PowerPoint PPT Presentation

Visualization + Analysis

Blockchains Are • Networks • Time-series

Visualization • Quickly spot properties • Quickly spot inconsistencies • Ask better questions

This Lecture • Analyze some properties of cryptocurrencies • Tools • Data Sources • Insights • Sample code

Distributions

Distributions • Distributions of: • Transaction Fees • Wallet net worths • Bitcoin Script Usage • Whales

Bitcoin Transaction Fees • BTC • Satoshi per byte • 100mn Satoshis = 1 BTC

Ethereum Transaction Fees • ETH • Gas • 21000 Gas = Base fee • Just transferring funds • Put down Gas Price • Pay Gas Price * Gas Used • Put down Gas limit

Bitcoin Transaction Fees Fee Txns 0 8380 1 9407071 2 2841101 … ….

Why Zeros? • Possibly: • Miners’ own transactions • Incredible generosity • Off-chain payment

Most Common Fee Fee Txns 1 9407071 3 7448408 4 2863087 … ….

Distribution • Raw bar chart bad for viz (large variance). • Solution: • log/log plot

Log-Log Plot

Log-Log Plot • Seems like a truncated power law

Power Laws • 80/20 Rule • Internet Networks • Traffic Arrival Times • Zipf • Twitter followers

Ethereum Gas Prices

log-normal?

Ethereum Gas Prices Fee Txns 0 121023 1 117 2 7 … ….

Most Common Fee Fee Txns 20000000000 11107198 1000000000 7354494 10000000000 7339890 … ….

Zero Gas Prices? • https://www.reddit.com/r/ethereum/comments/7lx1do/ a_christmas_mystery_sweepers_and_zero_gas_price/ • https://medium.com/chainsecurity/zero-gas-price-transactions- what-they-do-who-creates-them-and-why-they-might- impact-scalability-aeb6487b8bb0

Time-Series

Correlate With Events • Correlation to fiat? • Correlation to other coins?

Other Coins

YTD

Spearman’s Rank Correlation Col1 Col2

Spearman’s Rank Correlation Col1 Col2 Rank=2 Rank=3 Rank=10 Rank=5 Rank=5 Rank=2 Rank=3 Rank=1

Spearman’s Rank Correlation • Row-wise difference squared : d^2 • Sum up these row-wise differences •

Spearman’s Rank Correlation • +1/-1 : Strong positive / Strong negative • 0 : No correlation

Correlation Charts - Coins

Correlation Charts - BTC v. Fiat

So? • Looks like there is almost no correlation to fiat • Coins almost all move in lock-step • Implications?

BTC Volume Events

Network

Visualizing Networks • Slightly complex with bitcoin • The Bitcoin graph: • Nodes: wallet addresses • Edges: Spends

Visualizing Networks • Best practices contribute 1 or 2 nodes each transaction • In practice this seems to be 50%

Degree Distributions • Seems to be power law:

Ethereum

Transaction Patterns • Fork-merge: • Large amount in wallet • Split into many smaller wallets • Finally after a long trip merged into single wallet • Binary tree-like structure: • Transaction + Change • Splitting your amount into 2 • Long Chains

Insights From Degree • What about degree 1: • Likely money transferred to same individual • Large outdegree: • Possibly automated transaction

GeoSpatial

Hard Because • Many won’t expose an IP address • Many won’t respond to API calls that identify their address • Not very trustworthy

Visit At • https://blockchaincourse.onai.com/node_viz/

Questions?